[Comments or questions]
Copyright © 1996-2001 jsd
18 Stalls and Spins
Caution: Cape does not enable user to fly.
- warning label on Superman costume sold at Walmart
Spins are tricky. After reading several aerodynamics
texts and hundreds of pages of NASA spin-tunnel research reports,
I find it striking how much remains unknown about what happens
in a spin.
18.1 Stalls: Causes and EffectsHere's a basic yet important fact: if you don't stall the airplane, it
won't spin. Therefore, let's begin by reviewing stalls.
As discussed in section 5.3, the stall occurs at the
critical angle of attack, which is defined to be the point where a
further increase in angle of attack does not produce a further
increase in coefficient of lift.
Nothing magical happens at the critical angle of attack. Lift does
not go to zero; indeed the coefficient of lift is at its maximum
there. Vertical damping goes smoothly through
zero as the airplane goes through the critical angle of attack, and
roll damping goes through zero shortly thereafter. An
flying 0.1 degree beyond the critical angle of attack will behave
itself only very slightly worse than it would 0.1 degree below.
If we go far beyond the critical angle of attack
(the ``deeply stalled'' regime) the coefficient of lift
is greatly reduced, and the coefficient of drag is greatly increased.
The airplane will descend rapidly, perhaps at thousands of feet
per minute. Remember, though: the wing is still supporting
the weight of the airplane. If it were not, then there would
be an unbalanced vertical force, and by Newton's law the airplane
would be not only descending but accelerating downward.
If the wings were really producing zero force (for instance,
if you snapped the wings off the airplane) the fuselage would
accelerate downward until it reached a vertical velocity (several
hundred knots) such that weight was balanced by fuselage drag.
18.2 Stalling Part vs. All of the WingWe can arbitrarily divide the wing into sections; each section
contributes something to the lift of the whole wing. It is highly
desirable (as discussed in section 5.4.3) to have the
coefficient of lift for sections near the wing-root reach its maximum
early, and start decreasing, while the coefficient of lift for
sections near the tips continues increasing1 (as a
function of angle of attack).
Therefore it makes perfect sense to say that the
sections near the roots are stalled while the sections near the
tips are not stalled. If only a small region near the root is
stalled, the wing as a whole will still have an increasing
coefficient of lift — and will therefore not be stalled.
We see that the wing will continue to produce lots
of lift well beyond the point where part of it is stalling. This
is the extreme slow-flight regime — you can fly around all day
with half of each wing stalled (although it takes a bit of skill
and might overheat the engine).
18.3 Boundary LayersThere is a very simple rule in aerodynamics that says the velocity of
the fluid right next to the wing (or any other surface) is zero.
This is called the no-slip boundary condition. Next to the
surface there is a thin layer, called the
boundary layer, in which the velocity increases from zero to its
18.3.1 Separated versus Attached FlowThe wing works best when the airflow is attached to the wing surface by a simple boundary layer. The
opposite of attached flow is separated flow.
For attached flow, as we move through the boundary layer from the wing
surface out to the full-speed flow, there is practically no pressure
change. Sometimes it helps to think about attached flow in the
following way: Imagine removing the boundary layer and replacing it
with a layer of putty that redefines the shape of the wing. Then
imagine ``lubricating'' the new wing so that the air slides freely
past it; the no-slip condition no longer applies. Bernoulli's principle
can be used to calculate the pressure on the surface of the putty;
obviously it could never be applied inside the boundary layer. The
putty-covered wing may not be the most desirable shape, but it won't
necessarily be terrible.
For separated flow, the putty model does not work. Suppose I want to
pick up a piece of lint from the floor using a high-powered vacuum
cleaner. If I keep the hose 3 feet away from the floor, it will never
work; I could have absolute zero pressure at the mouth of the hose,
but the low pressure region would be ``separated'' from the floor and
the lint. If I move the hose closer to the floor, eventually it will
develop low pressure near the floor. This is part of the problem with
separated flow: there is low pressure somewhere, but not where you
need it. Separation can have multiple evil effects:
Separation means the air doesn't follow the contour of the
wing. This is somewhat like having a really thick boundary layer.
The wing can't force the air into the optimal flow pattern, so not as
much low pressure is produced.
- Whatever low pressure is produced isn't all attached to
the wing surface. This is a new problem that an attached flow would
not have, no matter how thick the boundary layer.
- On a non-streamlined object such as a golf ball,
there is a lot of drag (specifically: form drag, as discussed
in section 4.3) because separation disrupts a desirable
high-pressure area behind the ball.
18.3.2 Laminar versus Turbulent FlowIn the simplest case, there is laminar flow, in which every
small parcel of air has a definite velocity, and the velocity varies
smoothly from place to place. The other possibility is called turbulent flow, in which:
The closer we look, the more fluctuations we see.
at any given point the velocity fluctuates as a function of
- at any given time the velocity changes rapidly as we move from
point to point, even for nearby points.
Attached turbulent flow produces a lot of mixing. Some bits of air
move up, down, left, right, faster, and slower relative to
the average rearward flow.
For separated laminar flow, there will be some reverse flow (noseward,
opposing the overall rearward flow) but the pattern in space will be
much smoother than it would be for turbulent flow, and it will not
fluctuate in time.
You can tell whether a situation is likely to be turbulent if you know
the Reynolds number. You don't need to know the details, but
roughly speaking small objects moving slowly through
viscous fluids (like honey) have low Reynolds
numbers, while large objects moving quickly through thin fluids (like
air) have high Reynolds numbers. Any system with a Reynolds number
less than about 10 is expected to have laminar flow everywhere. If you
drop your FAA ``Pilot Proficiency Award'' wings into a jar of honey,
they will settle to the bottom very slowly. The flow will be laminar
everywhere, since the Reynolds is slightly less than 1. There will be
no separation, no turbulence, and no form drag — just lots of
Systems with Reynolds numbers greater than 10 or so are expected to
create at least some turbulence. Airplanes operate at Reynolds
numbers in the millions. The wing will have a laminar boundary layer
near the leading edge, but as the air moves back over the wing, at
some point the boundary layer will become turbulent. This is called
the transition to turbulence or simply the boundary layer
transition. Also at some point (before or
after the transition to turbulence) the airflow will become separated.
The designers try to keep the region of separation rather small and
near the trailing edge. In order to make a wing develop a lot of lift
without stalling, it helps to minimize the amount of separation.
18.3.3 Boundary Layer ControlOne scheme2 for controlling
separation involves the use of vortex generators (VGs). The VGs
are the little blades you see on the top of some wings, sticking up
into the airstream at funny angles. Each blade works like a turnplow,
reaching out into the high-velocity airstream and turning the layers
over — plowing energy into the inner
Re-energizing the boundary layer allows the wing to fly at higher
angles of attack (and therefore higher coefficients of lift) without
stalling. This improves your ability to operate out
of short and/or obstructed fields.
The vorticity created by these little VGs should not be confused with
the bound vortex, the big vortex that generates
the circulation that supports the weight of the airplane. As
discussed in section 3.12, to create lift you have to make
the air circulate around the wing; that is, there must be vortex
line running along the span. VGs don't do that; their vortex lines
run chordwise, not spanwise.3
Boundary-layer turbulence (whether created by VGs
or otherwise) also helps prevent separation, once again by stirring
additional energy into the inner sublayers of the boundary layer.
On a golf ball, 99% of the drag is form drag, and
only 1% is skin-friction drag. The dimples in the golf ball provoke
turbulence, adding energy to the boundary layer. This allows
the flow to stay attached longer, maintaining the high-pressure
region behind the ball, thereby decreasing the amount of form
drag. The turbulence of course increases the amount of skin-friction
drag, but it is worth it.4
Bernoulli's principle does
not apply inside the boundary layer, separated
or otherwise. As discussed in section 3.4, Bernoulli's
principle applies in situations where pressure (potential energy)
and airspeed squared (kinetic energy) add up to a constant. This
is not the case in the boundary layer, because friction there
converts a significant amount of the energy into heat.
Do VGs play the same role as dimples on a golf ball?
Not exactly. Unlike a golf ball, a wing is supposed to produce
lift. Also unlike a golf ball, a wing is highly streamlined;
consequently, its form drag is not predominant over skin-friction
drag. VGs are typically used to improve lift at high angles of
attack (by fending off loss of lift due to separation). They
may or may not improve performance at low angle of attack (by
decreasing form drag at the expense of skin-friction drag).
If you want ultra-low drag, and don't care about
short-field performance, you want a wing with as much laminar
flow as possible. Designing a ``laminar flow wing'' is
exquisitely difficult, especially in the real world where the
laminar flow could be disturbed by rain, ice, mud, and splattered
bugs on the leading edge.
There is always some separation on every airfoil
section. The separation grows as the angle of attack increases.
If there is too much separation, it cuts into the wing's ability
to produce lift. If there were no separation, the wing could
continue producing lift up to very high angles of attack (thereby
achieving very high coefficients of lift).
Having lots of separation is the dominant cause (but
not the definition) of stalling.5 Remember:
the stall occurs at the critical angle of attack, i.e. the point
where max coefficient of lift is attained.
A full discussion of turbulence and/or separated flow is beyond the scope
of this book; indeed, trying to really understand and control
these phenomena is a topic of current research. There is nothing simple
about it. But there are a few things we can say.
For more information, see e.g. reference 17.
- The opposite of separated flow is attached
- The opposite of turbulent flow is laminar
- Separated flow need not be very turbulent, nor
- Laminar flow need not be attached, nor vice versa.
- Turbulence doesn't cause separation (and indeed
oftenhelps prevent it).
18.4 Coanda Effect, etc.The name Coanda effect is generally applied to any situation
where a thin, high-speed jet of fluid meets a solid surface and
follows the surface around a curve. Depending on the situation, one or
more of several different physical processes might be involved in
making the jet follow the surface.
As a pilot, you absolutely do not need to know about the Coanda effect
or what causes it. Indeed, many professional aerodynamicists get along
just fine without really understanding such things. The main purpose
of this section is to dispel the notion that a normal wing produces
lift ``because'' of some type of Coanda effect.
Using the Coanda effect to explain the operation of a normal wing
makes about as much sense as using bowling to explain walking. To be
sure, bowling and walking use some of the same muscle groups, and both
at some level depend on Newton's laws, but if you don't already know how
to walk you won't learn much by considering the additional complexity
of the bowling situation.
18.4.1 Tissue-paper DemonstrationYou can demonstrate one type of Coanda effect for yourself using a
piece of paper. Limp paper, such as tissue paper, works better than
stiff paper. Drape the paper over your fingers, and then blow
horizontally, as shown in the following figures.6
If the jet passes just above the paper, as shown in figure 18.1, nothing very interesting happens. The jet just
keeps on going. The paper is undisturbed.
On the other hand, if the jet actually hits the paper as shown at
point C in figure 18.2, the downstream part of the paper will
rise up. This is because the air follows the curved surface; as it
does so, it creates enough low pressure to lift the weight of the
The air in your lungs, at point A, is at a pressure somewhat above
atmospheric. At point B, after emerging from the nozzle, the air in
the jet is at atmospheric pressure.
As discussed in section 3.3, the fact that the
fluid follows a curved path proves that there is a force on it. This
force must be due to a pressure difference. In this case, the
pressure on the lower edge of the jet (where it follows the curve of
the tissue paper near point D) is less than atmospheric, while the
pressure on the upper edge of the jet (near point E) remains
more-or-less atmospheric. This pressure difference pulls down on the
jet, making it curve. By the same token it also pulls up on the
paper, creating lift.
People who only half-understand Bernoulli's principle will be
surprised to hear that the jet leaves the nozzle at high speed at
atmospheric pressure. It's true, though. In particular, the crude
statement that ``high velocity means low pressure'' is an
oversimplification that cannot be used in this situation. The correct
basis of Bernoulli's principle is that for a particular parcel of
air the mechanical energy
(pressure plus kinetic energy per unit
volume) remains more-or-less constant. If you want to compare two
different parcels of air, you'd better make sure that they started out
with the same mechanical energy.
In this case, the air in the jet
leaves the nozzle with a higher mechanical energy than the ambient
air. Your lung-muscles are the source of the extra energy.
When this high-velocity, atmospheric-pressure air smacks into the
paper at point C, it actually creates above-atmospheric pressure
there. Indeed, we can use the streamline-curvature argument again: if
the air turns a sharp corner, there must be a very large pressure
In order to make this sharp turn, the air needs something to push
against. A good bit of the required momentum comes from the air that
splatters backward, as suggested by the squiggles just below and
upstream of the point of contact. This process is extremely messy.
It is much more complicated than anything that happens near a
wing in normal flight. To visualize this splatter, blow a jet of air
onto a dusty surface.7 Even if you blow at a very low
angle, some of the dust particles blow away in the direction
opposite to the main flow.
18.4.2 Blowing the Boundary LayerSince we saw in section 18.3 that de-energizing the
boundary layer is bad, you might think adding energy to the boundary
layer should be good... and indeed it is. One way of doing so uses
vortex generators, as discussed in section 18.3.
Figure 18.3 shows an even more direct approach.
We use a pump to create a supply of air at very high
- The air comes out a nozzle. The result is a jet of
high-velocity air at the same pressure as the local
- The jet shoots out of a slot in the top of the wing, adding
energy to the boundary layer at a place where this could be
Once again, the Coanda effect cannot explain how the wing works; you
have to understand how the wing works before you consider the added
complexity of the blower.
In this case we expect one spectacular added complexity, namely
curvature-enhanced turbulent mixing. This phenomenon will not be discussed in this book,
except to say that it does not occur near a normal wing, while it is
likely to be quite significant in the situation shown in figure 18.3.
Curving flows with lots of shear can be put to a number of other
fascinating uses, but a discussion is beyond the scope of this
book. See reference 9.
18.4.3 Teaspoon DemonstrationAnother example of a jet following a curved surface uses a jet of
water. You can easily perform the following experiment: let a thin
stream of water come out of the kitchen faucet. Then touch the left
side of the stream with the convex back side of a spoon. The stream
will not be pushed to the right, but instead it will follow the curve
of the spoon and be pulled to the left. The stream can be deflected
by quite a large amount. In accordance with Newton's third law of
motion, the spoon will be pulled to the right.
I don't understand everything I know about this situation, but it is
safe to say the following:
To convince yourself of these facts, it helps to have a higher
velocity and/or a larger diameter than you can conveniently get from a
kitchen faucet. A garden hose will give you a bigger diameter, and if
you add a nozzle you can get a higher velocity. You can easily
This water-in-air jet differs in fundamental ways from the
air-in-air jet situation described above.
- This effect has practically nothing to do with the way a normal
wing produces lift.
It appears that surface tension plays two very important roles:
The amount of lift9 you can produce is pathetically small, compared to the
dynamic pressure and area of the water jet.
- The lift-to-drag ratio is terrible. Indeed this makes
it very hard to measure the lift; if you get the angle slightly wrong
you will inadvertently measure a drag component instead.
- The water spreads out when it hits the surface, making a thin
coating over a wide area of the surface. This is in marked contrast
to what happens in the air-in-air jet, as you can demonstrate by
placing thin strips of tissue paper side by side. You can easily blow
on one strip and lift it without disturbing its neighbors.
- Some of the spreading layer flows backwards, ahead of the point
of contact of the jet, corresponding to a negative amount of upwash.
This is grossly different from what happens near a real wing.
- The effect does not depend on curvature-enhanced turbulent
mixing with the ambient air. This is quite unlike what happens in a
real airplane with boundary-layer blowing.
In both respects this is quite unlike the air-in-air jet, where the
air/wing surface tension has no effect and there is no such thing as
air/air surface tension.
At the water/air interface it prevents mixing of the air and
- At the water/wing interface it plays a dominant role in making the
water stick to the surface.
To convince yourself of this: Take a thin sheet of plastic. Get it
wet on both sides, and drape it over a cylinder. You will not be able
to lift it off the cylinder using a tangential water jet. The surface
tension holding the wet plastic to the cylinder is just as strong as
the tension between the plastic and the jet. In contrast, when the
same piece of plastic has air on both sides, you can easily lift it
off the cylinder using an air jet.
18.4.4 Fallacious Model of Lift ProductionYou may have heard stories saying that the Coanda effect explains how
a wing works. Alas, these are just fairy tales. They are worse than
Don't let anybody tell you that squirting a spoon or blowing on tissue
paper is a good model of how a wing works.
For starters, these fairy tales often claim that blowing on
tissue paper (as described just above) proves that ``high
means low pressure'' which is absolutely not what is being
demonstrated. The high-velocity air coming out of your mouth is at
atmospheric pressure. If you blow across the top of a flat
piece of paper, it will not rise, no matter what you do. There is no
low pressure in the jet (unless and until it gets pulled around a
corner). Therefore the Coanda stories give a wrong explanation of
normal wings and basic aerodynamics. And by the way, such stories
cannot even begin to explain the operation of flat wings — yet we
have seen in section 3.10.1 that a barn door
doesn't behave very differently from other airfoils.
- The Coanda-like notion of airflow following a curved surface
cannot possibly explain why there is upwash in front of the wing. In
figure 18.2 there must be a stagnation point on the upper
surface of the paper near point C. This is completely different
from the situation near a normal wing, where the stagnation line must
be somewhere below the leading edge of the wing. Upwash
is important, since it contributes to lift
while creating a negative amount of induced drag. A further
consequence, by the way, is that these Coanda-like stories cannot
possibly explain the operation of stall-warning devices, as discussed
in section 3.5.
- As mentioned above, the distribution of velocities necessary
to create curvature-enhanced turbulent mixing is produced by a
high-speed jet but is not produced by a normal wing.
- Sometimes the fairy tales say that the jet ``sticks'' to the surface
because of viscosity. This implies that if the viscosity of the
fluid changes, the amount of lift an airfoil produces should change in
proportion. In fact, though, the amount of lift produced by a real
wing is independent of viscosity over a wide range. Also, many of the
processes responsible for the real Coanda effect require the
production of turbulence, so they only work if the viscosity is
- In the real Coanda effect, we know where the high-velocity
air comes from. It comes from a nozzle. Upstream of the nozzle is a
pump (or a rocket engine, or some other device) to supply the
necessary energy. The jet makes high-velocity air above the wing, not
below, because that's where we aim the nozzle. An ordinary wing is
completely different. It is wonderfully effective at creating
high-velocity air above itself, without nozzles, without pumps, and
without transferring energy11 to the air.
- The fairy tales generally neglect the fact that the wing speeds up the
air in its vicinity, and just assume that the relative wind meets the
wing at the free-stream velocity and follows the curve in a
Coanda-like way. As a consequence, they miscalculate the pressure
gradients by a factor of ten or so.
- Finally, in the real Coanda effect we know how big the jet is. Its
initial size is determined by the nozzle. The amount of mixing
depends on the speed of the jet, the speed of the ambient air, the
curvature of the surface, and other known quantities. Awareness of
the Coanda effect is a small part of — not a replacement for — a
full analysis of the wing in figure 18.3.
In contrast, (a) the typical fairy tales imply that the entire flow
pattern of a normal wing can be explained by mentioning the magic
words ``Coanda effect'', yet (b) they cannot explain how thick a
chunk of air is deflected by the wing. One inch? Six inches? A
chord-length? A span-length? Some amount proportional to the
viscosity of the air?
It would be very hard to calculate how
If you want to ``get the feel'' of lift production, the obvious
methods are the best. These include holding a model
airfoil13 downstream of a
household fan, or sticking
it out of a car
18.5 Spin EntryCase 1: In normal flight, rolling motions are very
heavily damped, as discussed in section 5.4. Even
though the static stability of the bank angle is small
or even negative, you cannot get a large roll rate without
a large roll-inducing force; when you take away the force the
roll rate goes away.
Case 2: Near the critical angle of attack, the roll
damping goes away. Suppose you start the aircraft rolling to the
right. The roll rate will just continue all by itself. The right
wing will be stalled (beyond max lift angle of attack) and the
left wing will be unstalled (below max lift angle of attack).
Case 3: At a sufficiently high initial angle of attack (somewhat
greater than the critical angle of attack), the roll will not just
continue but accelerate, all by itself. This is an example of the
constitutes the beginning of a snap roll or spin.
The resulting undamped rolling motion is called autorotation.
At a high enough angle of attack, the ailerons lose
effectiveness, and at some point they start working in
reverse.15 Figure 18.4 shows
how this reversal occurs. Suppose you deflect the ailerons
to the left. This raises the angle of attack at the right wingtip
and lowers it at the left wingtip. Normally, this would increase
the lift on the right wing (and lower it on the left), creating
a rolling moment toward the left. Near the critical angle of
attack, though (as seen in the left panel of the figure), raising
or lowering the angle of attack has about the same effect on the
coefficient of lift, so no rolling moment is produced (for now,
We see that at this angle of attack, anything that creates a rolling
mo-ment will cause the aircraft to roll like crazy, and indeed to keep
accelerating around the roll axis. There will be no natural roll
damping, and you will be unable to oppose the roll with the ailerons.
There are two main ways of provoking a spin at this point:
Suppose the airplane is in a steady slip to the left. That is, you
are steadily pushing on the right rudder pedal. Then the slip/roll
cou-pling (as discussed in section 9.1 and
section 9.2) will cause it to spin to the right.
- Suppose the airplane is not in much of a slip, but you suddenly
cause it to yaw to the right. The left wingtip will temporarily be
moving faster, and the right wingtip will temporarily be moving
slower. This difference in airspeeds will create a difference in
lift, causing a spin to the right. The initial yawing motion could
come from a sudden application of rudder, or from adverse yaw, or
what-ever. Note that in the right panel of
figure 18.4, the ai-leron deflection has a tremendous
effect on the drag. This means that ailerons deflected to the left
cause a yaw to the right which in turn provokes a roll to the just
the opposite of what ailerons normally do.
18.6 Types of Spin
The word ``spin'' can be used in several
different ways, which we will discuss below.
The spin family tree includes:
Figure 18.5 shows an airplane in a steady
spin. You can see that the direction of flight has two components:
a vertical component (down, parallel to the spin axis) and a horizontal
component (forward and around).
- ``departure'', i.e. onset of undamped
- incipient spin — i.e. one that has just gotten
- well-developed spin, which could be
a steep spin, or
- a flat spin.
Figure 18.6 is a close-up of a wing
in a steep spin. We have welded a pointer to each wingtip, indicating
the direction from which the relative wind would come if the wing
were producing zero lift; we call this the Zero-Lift Direction
(ZLD). (For a symmetric airfoil, the ZLD would be aligned with
the chord line of the wing.) Remember that
the angle between the direction of flight and the ZLD pointer
is the angle of attack.
In this situation, both wingtips have the same vertical
speed, but they have significantly different horizontal speeds
— because of the rotation. Consequently they have different directions
of flight, as shown in the figure. This in turn means that the
two wingtips have significantly different angles of attack, as
shown in figure 18.7. The two wings are producing
equal amounts of lift, even though one is in the stalled regime
and one in the unstalled regime.
Figure 18.8 shows another spin mode.
This time the rotation rate is higher than previously. The spin
axis is very close to the right wingtip. The outside wing is
still unstalled, while the inside wing is very, very deeply stalled,
as shown in figure 18.9.
: Doubly Stalled Flat
Spin — Coefficient of Lift
Figure 18.10 shows yet another possible spin mode. In
this case, the outside wing is stalled, while the inside wing is, of
course, much more deeply stalled. Whether this spin mode, or the one
shown in figure 18.9 (or both or neither) is stable
depends on dozens of details (aircraft shape, weight distribution, et
cetera). There is a common misconception that in a spin, one wing is
stalled and the other wing is always unstalled. This is true of
``most'' spins but it is not a defining property. It would be safer
to use the following definition:
In a spin, at least one wing is stalled,
and the two wings are operating at
very different angles of attack.
18.6.2 Samaras, Flat Spins, and Centrifugal ForceA samara is a winged seed. Maples are a particularly
well known and interesting example.
Maple samaras have only one wing, with the seed all
the way at one end. Its mode of flight is analogous to an airplane
in a flat spin. In an airplane, the inside wing is deeply stalled,
while in the samara the inside wing is missing entirely.
In a non-spinning airplane, if one wing were producing
more lift than the other, that wing would rise. So the question
is, why is a flat spin stable? Why doesn't the outside wing continue
to roll to ever-higher bank angles? The secret is centrifugal
you hold a broomstick by one end while you spin around and around;
the broomstick will be centrifuged outward and toward the horizontal.
In an airplane spinning about a vertical axis, the
high (outside) wing will be centrifuged outward and downward (toward
the horizontal), while the low (inside) wing will be centrifuged
outward and upward (again toward the horizontal). In a steady
flat spin, these centrifugal forces cancel the rolling moment
that results from one wing producing a lot more lift than the
other. This is the only example I can imagine where an airplane
is in a steady regime of flight but one wing is producing more
lift than the other.
As discussed in reference 6, an aircraft
with a lot of mass in the wings will have a stronger centrifugal
force than one with all the mass near the centerline of the fuselage.
In particular, an aircraft with one pilot and lots of fuel in
the wing tanks could have completely different spin characteristics
than the same aircraft with two pilots and less fuel aboard.
18.6.3 NASA Spin StudiesIn the 1970s, NASA conduced a series of experiments
on the spin behavior of general-aviation aircraft; see reference 8 and reference 7 and other papers cited therein.
They noted that there was ``considerable confusion'' surrounding the
definition of steep versus flat spin modes, and
offered the classification scheme shown in table 18.1.
The angle of attack that appears in this table is measured in
the aircraft's plane of symmetry; the actual angle of attack at
other positions along the span will depend on position.
|angle of attack
||20 to 30
||30 to 45
||45 to 65
||65 to 90
|rate of descent
|rate of roll
|rate of yaw
wingtip-to-wingtip difference in angle of attack
|nose-to-tail difference in slip
The NASA tests demonstrated that general aviation aircraft not
approved for intentional spins commonly had unrecoverable flat
18.6.4 Effects of Spin-Axis ChangesIn all cases NASA studied, the flat spin had a faster
rate of rotation (and a slower rate of descent) than the steep
spin. Meanwhile, reference 15 reports experiments in
which the flatter pitch attitudes were associated with the slower
rates of rotation. This is not a contradiction, because the latter
dealt with an unsteady spin (with frequent changes in pitch attitude),
rather than a fully stabilized flat spin. A sudden change
to a flatter pitch attitude will cause a temporary reduction
in spin rate, for the following reason.
In any system where angular momentum is not
the system will spin faster when the mass is more concentrated
near the axis of rotation. The general concept is discussed in
section 19.8. By the same token, if the mass
of a spinning object is redistributed farther from the axis, the
rotation will slow down.
When the spinning airplane pitches up into a flatter
attitude, whatever mass is in the nose and tail will move farther
from the axis of rotation. Angular momentum doesn't change in
the short run, so the rotation will slow down in the short run.
In the longer run — in a steady flat spin — the aerodynamics
of the spin will pump more angular momentum into the system, and
the rotation rate will increase quite a lot. The rotation rate
of the established flat spin is typically twice that of the steep
Recovering from an established flat spin requires
forcing the nose down. This brings the mass in the nose and tail
closer to the axis of rotation. Once again using the principle
of conservation of angular momentum, you can see that the rotation
rate will increase (at least in the short run) as you do so —
which can be disconcerting.
18.7 Recovering from a SpinIf you find yourself in an unusual turning, descending
situation, the first thing to do is decide whether you are in
a spiral dive or in a spin. In a spiral dive, the airspeed will
be high and increasing; in a spin the airspeed will be low. You
should be able to hear the difference. Also, the rate
of rotation in a spiral is much less; the high speed means the
airplane has lots of momentum and can't turn on a dime. In a
spin, the aircraft will be turning a couple hundred degrees per
To get out of a spin,17
follow the spin-recovery procedures given in the Pilot's Operating
Handbook for your airplane. The literature is full of home-brew
spin recovery procedures that probably work most of the time in
most airplanes, but if you want a procedure that works for sure,
follow the handbook for your airplane.
For typical airplanes, the spin recovery procedure
contains the following items:
Now let's discuss each of these items in a little
- Retard the throttle to idle
- Retract the flaps
- Neutralize the ailerons
- Apply full rudder in the direction opposing the
- Briskly move the yoke to select zero angle of
Retarding the throttle is a moderately good idea for a couple of
reasons. For one thing (especially if you have a fixed-pitch prop) it
keeps the engine from overspeeding during the later stages of the spin
recovery. More importantly, gyroscopic precession of the rotating
engine and propeller can hold the nose up, flattening the spin and
interfering with the recovery (depending on the direction of spin).
Propwash might increase the effectiveness of the
horizontal tail and therefore assist in the spin recovery,
but (especially in a flat spin) the propwash could be blown somewhere
else by the abnormal airflow — so you may not be able to count
Retracting the flaps is a moderately good idea because you might
exceed the ``max flaps-extended speed'' if you mishandle the later stages of the spin
recovery and you don't want to damage the flaps.
Retracting the flaps may help with the spin recovery itself. Recall
from section 5.4.3 that the flaps effectively increase the
washout of the wings. Washout ensures that the airplane will stall
before it runs out of roll damping. (This produces a nice
straight-ahead stall.) In the spin, though, when you have lost all
vertical damping and roll damping, the washout doesn't help.
The early stages of spin recovery are not like the early stages of
Neutralizing the ailerons is usually a good idea
for the simple reason that it is hard to think of anything better
to do with them. Deflecting the ailerons effectively increases
the angle of attack of one wingtip and decreases the angle of
attack of the other wingtip. In a spin, the part of the wing where
the ailerons are may (or may not) be in the stalled regime — so
deflecting the ailerons to the left may (or may not) produce a
paradoxical rolling moment to the right.
Depressing the rudder to oppose the spin is obviously
a good thing to do.
Finally, you want to move the yoke to select zero angle of
attack. In typical trainers,
this means shoving the yoke all the way forward, but in other
aircraft, especially aerobatic aircraft, all the way forward might
select a large negative angle of attack. Shoving the yoke all
the way forward in such a plane would likely convert the spin to an
inverted spin — hardly an improvement. This is just one example of
why you want to know and follow the spin recovery procedure for your
The relative significance of the rudder compared
to the flippers in breaking the spin depends radically on the
design of the airplane, the loading of the airplane, and on the
spin mode, as discussed in reference 6.
In normal non-spinning flight, you should apply smooth
pressures to the controls. Spin recovery is the exception: it
calls for brisk, mechanical
motions of the controls, almost
without regard to the pressures involved.
If you get into a spin in instrument conditions,
you should rely primarily on the airspeed indicator and the rate-of-turn
gyro. The inclinometer ball cannot be trusted; it is likely to
be centrifuged away from the center of the airplane — giving an
indication that depends on where the instrument is installed,
telling you nothing about the direction of spin. The artificial
horizon (attitude indicator) cannot be trusted since it may have
tumbled. The rate-of-turn gyro is more
trustworthy, since it
is a rate gyro, not a free gyro; that is, it has no gimbals and
cannot possibly suffer from gimbal lock.
Recovery from a so-called incipient spin (one that
has just gotten started) is easier than from a well-developed
spin. Normal-category single-engine18 certification requirements say that
an airplane must be able to recover from a one-turn spin (or a
3-second spin, whichever takes longer) in not more than one additional
turn. If you let the spin go on for several turns, you might
progress from a steep spin to a flat spin. Recovery could take
a lot longer — if it is possible at all.
If you load the airplane beyond the aft limit of the weight and
balance envelope, even the incipient spin may be unrecoverable; see
section 6.1.1. Imperfect repairs to the wing, or slack in
the control cables, could also impede spin recovery.
Finally, the spin is yet another reason why it is
NOT SAFE to think of the yoke as simply the up/down
control.19 In a
spin you have a low airspeed and a high
rate of descent. If you think of the yoke as the up/down control,
you will be tempted to pull back on the yoke, which is exactly
the wrong thing to do. On the other hand, if you think of the
yoke as (primarily) the fast/slow control, you will realize that
you need to push forward on the yoke, to solve the airspeed problem.
18.8 Don't Mess With SpinsIt is quite impressive how well a samara works.
A maple seed descends very slowly, riding the wind much better
than a parachute of similar size and weight ever could. Flat
spins can be extremely stable; a wing by itself loves to
spin. That's why spins (and flat spins in particular) are so
dangerous: it takes a lot of rudder force to persuade a wing to
Spins are extremely complex. Even designers and
top-notch test pilots are routinely surprised by the behavior
of spinning airplanes. Spin-test airplanes are equipped with
cannon-powered spin-recovery parachutes on the airframe, and quick-release
doors in view of the distinct possibility that the pilot will
have to bail out. Tests are conducted at high altitude over absolutely
unpopulated areas. Therefore please don't experiment with spinning
a plane except exactly as approved by the manufacturer — one unrecoverable
spin mode can ruin your whole day.
- This happens naturally on a rectangular wing; it can be
enhanced by washout and other designers' tricks.
- An even more direct method of
adding energy to the boundary layer uses a jet of high-velocity air,
as discussed in section 18.4.2.
- Of course, the
VGs contribute indirectly to maintaining the health of the big
bound vortex, since they help maintain attachment and therefore help
create lots of circulation.
- See reference 17 for a nice discussion of golf balls, cricket balls,
and boundary layers in general.
just as having lots of water is the cause, but not the definition,
of drowning — you can get very wet without drowning.
- You can blow directly from your lips, but it's better to
use a flexible straw or a thin piece of tubing, so that you can get a
better view of what's happening. If you put a nozzle at the end of
the tube, the jet will keep its shape better.
- Ground pepper is a
convenient source of suitable dust.
- This won't be exactly atmospheric,
since the local pressure has been affected by the wing.
- Remember, lift
is the force perpendicular to the flow and perpendicular to the
- Indeed, as long as
the viscosity is not exactly zero, the smaller the viscosity,
the greater the turbulence.
- Of course
some energy is transferred, in the form of friction and induced
drag, but this is very small, out of all proportion to the energy that
the air parcel transfers from its own speed to pressure and back
- Nonsensical things are often rather
hard to calculate.
- If you don't have a good model
airfoil, start with a flat piece of cardboard.
- This refers to ``departure from normal flight''. It
has nothing to do with takeoff or with a ``departure stall'' which
merely refers to a stall in the takeoff configuration.
- Under present-day certification rules, the ailerons
are required to work normally up to at least stalling
angle of attack. However, some older airplanes were built under
older rules. These planes, including many aerobatic aircraft,
have much less washout, and therefore lose aileron effectiveness
earlier. All planes lose effectiveness eventually. For simplicity,
this section ignores washout.
- See section 19.4
for a discussion of the nature of centrifugal fields.
from a spiral dive is discussed in section 6.2.4.
aircraft are not required to be recoverable from any sort of spin,
incipient or otherwise.
- This point is discussed in chapter 7.
[Comments or questions]
Copyright © 1996-2001 jsd